I wonder if it’s just me unable to find it, or if it doesn’t exist — an app for Mac where you can specify a folder with PDFs, let it work overnight to create a vector index, and then search through them using natural language queries. For creating embeddings and for the RAG reasoning phase, models could be downloaded onto the computer, and in this case, it would work as fast as the computer allows. If you enter an OpenAI key, then OpenAI would be utilized.
It would also be convenient if such a tool could take a URL as input – for example, through a click on a browser extension button, and then allow searches across all documents, including that URL.
It would be great if one could create indexes with a single click. Say you receive a 200-page PDF on biology, you throw it into the “biology” index, or one on mathematics, you throw it into the “mathematics” index.
I know how to build this from scratch. But all solutions require a somewhat advanced understanding. There’s something like automatic1111 for Stable Diffusion; I wish there was something similar for working with an archive.
Hi, I'm Rauf Aliev. 